CN103513249A - Broadband coherent mold base signal processing method and system - Google Patents

Abstract

The invention brings forward a broadband coherent mold base signal processing method and system. The method comprises the steps of 101) obtaining array reception data of B frequency points, performing wave beam normalization of data of full frequency bands and conducting in a coherent way to form a comprehensive reception data vector; 102) obtaining modeling vectors of the B frequency points and performing coherence of the modeling vectors of the full frequency bands to form s comprehensive expected guiding vector; 103) obtaining an optimum guiding vector based on the comprehensive reception data vector and the comprehensive expected guiding vector; and 104) obtaining a final target positioning result based on the optimum guiding vector. The step 102) includes traversing possible positions for appearance of sound sources, inputting a modeling tool to obtain weight vectors of the B frequency points and iterating the B weight vectors to finally obtain the comprehensive expected guiding vector. The step 103) includes processing the comprehensive reception data vector and the comprehensive expected guiding vector in a correlative manner and traversing all possible sound source positions to obtain a decision function, wherein the position of the maximum value of the decision function is the position of sound sources.

Description

The relevant mould base signal processing method in a kind of broadband and system

Technical field

The invention belongs to sonar digital processing field, particularly the relevant mould base signal processing method in a kind of broadband.

Background technology

In sonar system, according to the difference of principle of work, be divided into active sonar and passive sonar.Passive sonar itself is not launched acoustical signal, just accepts passively the radiated noise of suspicious object, to carry out target detection, then carries out location and the identification of target sound source.

Passive sonar auditory localization needs the information such as space, bandwidth, to obtain lower secondary lobe, and improves the robustness of environment mismatch and the inhibition ability to noise.Ocean propagation channel is frequency change, the decline of channel and time loose characteristic be that frequency dependence due to marine environment earthquake sounds attribute causes in itself.To fixing marine site, adopt single-frequency or narrow band signal to carry out mould base and process and may cause that resolution is low, robustness is poor and the problem such as uniqueness, and can obtain more excellent result by enough wide broadband signal.

The Wideband Signal Processing method can be divided into two kinds: a kind of is traditional non-coherent approaches, and another kind is coherent approach.Noncoherent disposal route has been utilized the spatial coherence in a frequency, but does not consider the coherence messages between frequency, and it is average that the ambiguity surface of a plurality of Frequency points directly carries out real number field.Current Processing Algorithm adopts the method for incoherent processing between this spatial coherence processing, frequency mostly.Relevant disposal route has been considered the simple crosscorrelation information between frequency, and the selection weighting coefficient relevant with frequency be weighted on average, due to the coherence who has considered between frequency, can utilize more fully underwater sound propagation characteristic, thus the accuracy that raising is located.The relevant research of processing in ，Dui broadband is at present a heat subject.。

The relevant mould base Processing Algorithm in initial broadband realizes in time domain; in document 1 " C.S.Clay.Optimum time domain signal transmission and source localization in a waveguide.J.Acoust.Soc.Am.; 81:660-664; 1987 "; having realized the earliest the relevant Matched Field in broadband of time domain processes; he proposes to obtain impulse response with the impulse response measuring and modeling and matches; with this, estimate sound source position, he also expands to the method the cross correlation process of many array elements.Document 2 " R.K.Brienzo, W.S.Hodgkiss.Broadband matched-field processing.J.Acoust.Soc.Am., 94:2821-2831,1993 " is used this algorithm successfully to locate in test the explosive sound source at 9km place.But this algorithm need to be known the frequency spectrum of sound source in advance, this is impossible under the application conditions of most passive sonar.In document 3 " P.Hursky; M.B.Porter; M.Siderius.High frequency (8-16hz) model-based source localization.J.Acoust.Soc.Am.; 115:3021-3032; 2004 ", the performance of having studied high band (more than 1kHz) time domain broadband MFP, because high band is more responsive for parameter perturbation, auditory localization is more difficult.First they suppose sound source known disturbance waveform, thereby the method for using matched filtering obtains the impulse response of channel, the impulse response obtaining with modeling matches, can obtain the estimation of sound source position, when waveform is unknown, the signal that they receive two nautical receiving sets is done cross correlation process, thereby has eliminated the impact of sound source waveform on processor, gives the tracking results of the sound source degree of depth and distance in literary composition.

Document 4 " E.K.Westwood.Broadband matched-field source localization.J.Acoust.Soc.Am.; 91 (5): 2777-2789; May 1992 " has been realized the relevant mould base in broadband at frequency domain and has been processed; with the relevant unity of thinking of processing of time domain broadband; in frequency domain, be also that relevant treatment is done in the impulse response that the impulse response measuring and modeling are obtained.Different from the method for time domain, processor output is no longer the maximal value of the related function of time domain, but the coherent accumulation of frequency domain simple crosscorrelation.Because this processor has been done relevant treatment at each frequency, so do not need to know the spectrum information of sound source.Document 4 is pointed out, the relevant processor in broadband is better than incoherent processor, in relevant processor, removing autocorrelative part (being CSDM Zhong diagonal angle item) effect can be better, in addition, he also discusses Liao Zhen aperture, the impact of signal bandwidth on processor performance, uses more array element, larger bandwidth ,Geng great aperture will obtain positioning result better.In literary composition, give the test findings at 5000m deep-sea, use vertical array at the bottom of 200 meters of anchors, successfully audio source tracking has been arrived to the distance of 42km.

Document 5 " Z.-H.Michalopoulou and M.B.Porter.Matched-field processing for broad-band source localization.IEEE Journal of Oceanic engineering, 21:384-392, 1996 " a kind of coherent signal subspace method (array element normalization broadband matched-field processors that is concerned with of direct coupling sound field has been proposed, be called for short MP algorithm), the method of the coupling simple crosscorrelation before being different from, this algorithm by array received to the sound field iterate of each frequency get up, form a super vector, due to the phase differential impact between frequency, directly sound field matching effect is very poor, they have proposed a kind of normalized method, the vector of each frequency be take first array element as with reference to normalization, then the vectorial iterate of each frequency is got up, form super vector, to eliminate the impact of phase differential between frequency, the method can be directly used in the adaptive algorithms such as MVDR.They are used for test figure by the relevant method in this broadband, and result shows that the relevant processor in MVDR broadband reaches 90% to the tracking of sound source distance and the degree of depth is efficient, and noncoherent MVDR is only 10%.But this method when low signal-to-noise ratio, normalized successful variation, thus can not eliminate the impact of sound source.

The technological means that technical scheme of the present invention adopts has effectively overcome the technological deficiency that document 5 exists.The disposal route of the mould base signal of prior art as shown in Figure 1.

Summary of the invention

The object of the invention is, the treatment loss bringing for overcoming the incoherent processing of existing wide-band-message, and the in the situation that of low signal-to-noise ratio, target detection ability declines, and proposes the relevant mould base disposal route in a kind of broadband and system.

Wave beam normalization of the present invention broadband mould base Processing Algorithm, the reference vector normalization after the array signal of each frequency is formed with wave beam, thereby the minimum detectable signal to noise ratio of reduction processor, improve and detect performance.

For achieving the above object, the invention provides the relevant mould base signal processing method in a kind of broadband, described method comprises:

Step 101) obtain the array received data of B frequency, the data of Whole frequency band are done to wave beam normalization, relevant formation comprehensively receives data vector;

Step 102) obtain the modeling vector of B frequency, and the comprehensive expectation of the relevant formation of the modeling vector steering vector to Whole frequency band;

Step 103) according to described comprehensive reception data vector and comprehensive expectation steering vector, obtain optimum steering vector;

Step 104) according to optimum steering vector, obtain final target localization result.

In technique scheme, described step 101) be specially: the array received data to B the frequency receiving are carried out Fast Fourier Transform (FFT), obtain the data of each single-frequency point; The data of B frequency are carried out to wave beam formation and normalized successively, form the comprehensive data vector that receives; Wherein, B value is for being more than or equal to 1.The array received data configuration array covariance matrix of L snap of accumulation.

In technique scheme, described step 101) further comprise following sub-step:

Step 101-1) by linear array, receive spacing wave, obtain the snap time-domain signal of N array element constantly;

Step 101-2) time domain data is done to Fast Fourier Transform (FFT), obtain the array data (X of B frequency ₁, X ₂..., X _b), wherein the array data Xi of i frequency is expressed as follows

X _i=[x _i1,x _i2,...,x _iN] ^T

X wherein _ijthe reception data that represent i frequency of j array element, T represents transposition;

Step 101-3) data of B frequency are done to wave beam and form, obtain the beam data X of i frequency _{i, beam}for

X _i,beam=X _i*e _i

Wherein, e _ithe array orientation compensation vector that represents i frequency, is determined by element position, target azimuth and signal frequency;

Reference signal (X using this beam data as each frequency _1ref, X _2ref..., X _bref),

X _iref=X _i,beam

Array data (X with this reference signal to each frequency ₁, X ₂..., X _b) do normalization, obtain the data after normalization

as shown in the formula:

$({\hat{X}}_1,{\hat{X}}_2,...,{\hat{X}}_B)=(\frac{X_1}{{\mathrm{<figure-callout\; id="1"\; label="X"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_{1\mathrm{ref}}},\frac{X_2}{{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2\mathrm{ref}}},...,\frac{X_B}{X_{\mathrm{Bref}}})$

Step 101-4) by the data after B frequency normalization

iterate gets up, the comprehensive reception data vector that to form a length be B*N

if note:

${\hat{X}}_1={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N}]}^T$

Wherein

represent the data after the normalization of i frequency of j array element, the normalized vector after expansion:

${\mathrm{<figure-callout\; id="1"\; label="X\; e"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_e={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N},{\hat{x}}_{21},{\hat{x}}_{22},...,{\hat{x}}_{2N},...,{\hat{x}}_{\mathrm{<figure-callout\; id="1"\; label="B"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">B</figure-callout>}1},{\hat{x}}_{\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">B</figure-callout>}2},...,{\hat{x}}_{\mathrm{BN}}]}^T;$

Step 101-5) receive L snap data constantly, respectively by step 101-2)-101-4) obtain the comprehensive reception data vector of each snap, the spread vector of whole L snaps is designated as X _{e, all};

X _e，all=[X _e1,X _e2,..X _eL]

Described step 101-1) be to receive snap data constantly, need to repeat L time step 101-5) mean this;

Step 101-6) use the comprehensive reception data vector of L snap to form array covariance matrix

$R=E\left[X_{e,\mathrm{all}}{X}_{e,\mathrm{all}}^H\right]---\left(1\right)$

Wherein, E represents to do statistical average, and H represents conjugate transpose.

In technique scheme, described step 102) concrete steps are: the position that traversal sound source may occur, and in conjunction with ocean environment parameter, input modeling tool obtains the weighing vector of B frequency, B weighing vector iterate got up, obtain comprehensively expecting steering vector.

In technique scheme, described step 103) further comprise:

Step 103-1) described comprehensive reception data vector and comprehensive expectation steering vector are done to the decision function value that relevant treatment obtains this sound source position, and travel through all possible sound source position, obtain about the decision function of whole sound source positions as follows:

P _coh-conv(θ，r，z)=w _e ^H(θ，r，z)Rw _e(θ，r，z)

The decision function P of step 103-2 sound source position _coh-convthe position that the maximal value of (θ, r, z) occurs is the estimated value of sound source position.

Based on said method the present invention, also provide a kind of broadband mould base signal processing system that is concerned with, described system comprises:

The comprehensive data vector acquisition module that receives, for obtaining the array received data of B frequency, does wave beam normalization to the data of Whole frequency band, and relevant formation comprehensively receives data vector;

Comprehensive expectation steering vector acquisition module, for obtaining the modeling vector of B frequency, and is concerned with and forms comprehensive expectation steering vector the modeling vector of Whole frequency band;

Overall treatment module, for the comprehensive reception data vector according to described and comprehensive expectation steering vector, obtains optimum steering vector; And obtain final target localization result according to optimum steering vector.

In technique scheme, described comprehensive reception data vector acquisition module carries out Fast Fourier Transform (FFT) to the array received data of B the frequency receiving, and obtains the data of each single-frequency point; The data of B frequency are carried out to wave beam formation and normalized successively, form the comprehensive data vector that receives; Accumulate L snap array received data configuration array covariance matrix.

In technique scheme, described comprehensive reception data vector acquisition module further comprises:

Receive submodule, for receiving spacing wave by linear array, obtain the snap time-domain signal of N array element constantly;

First processes submodule, and time domain data is done to Fast Fourier Transform (FFT), obtains the array data (X of B frequency ₁, X ₂..., X _b), wherein the array data Xi of i frequency is expressed as follows

X _i=[x _i1,x _i2,...,x _iN] ^T

X wherein _ijthe reception data that represent i frequency of j array element, T represents transposition;

Second processes submodule, for the data of B frequency being done to wave beam, forms, and obtains the beam data X of i frequency _{i, beam}for

X _i，beam=X _i*e _i

Wherein, e _ithe array orientation compensation vector that represents i frequency, is determined by element position, target azimuth and signal frequency;

Reference signal (X using this beam data as each frequency _1ref, X _2ref..., X _bref),

X _iref=X _i，beam

Array data (X with this reference signal to each frequency ₁, X ₂..., X _b) do normalization, obtain the data after normalization

as shown in the formula:

$({\hat{X}}_1,{\hat{X}}_2,...,{\hat{X}}_B)=(\frac{X_1}{{\mathrm{<figure-callout\; id="1"\; label="X"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_{1\mathrm{ref}}},\frac{X_2}{{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2\mathrm{ref}}},...,\frac{X_B}{X_{\mathrm{Bref}}})$

Iterate submodule, for by the data after B frequency normalization

iterate gets up, the comprehensive reception data vector that to form a length be B*N

if note:

${\hat{X}}_1={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N}]}^T$

Wherein

represent the data after the normalization of i frequency of j array element, the normalized vector after expansion:

${\mathrm{<figure-callout\; id="1"\; label="X\; e"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_e={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N},{\hat{x}}_{21},{\hat{x}}_{22},...,{\hat{x}}_{2N},...,{\hat{x}}_{\mathrm{<figure-callout\; id="1"\; label="B"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">B</figure-callout>}1},{\hat{x}}_{\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">B</figure-callout>}2},...,{\hat{x}}_{\mathrm{BN}}]}^T;$

The 3rd processes submodule, for receiving L snap data constantly, respectively by step 101-2)-101-4) obtain the comprehensive reception data vector of each snap, the comprehensive reception data vector of whole L snaps is designated as X _{e, all};

X _e，all=[X _e1,X _e2,...X _eL]

Described step 101-1) be to receive snap data constantly, need to repeat L time step 101-5) mean this;

Array covariance matrix forms submodule, for using the comprehensive reception data vector of L snap to form array covariance matrix

$R=E\left[X_{e,\mathrm{all}}{X}_{e,\mathrm{all}}^H\right]---\left(2\right)$

Wherein, E represents to do statistical average, and H represents conjugate transpose.

In technique scheme, the position that described comprehensive expectation steering vector acquisition module traversal sound source may occur, and in conjunction with ocean environment parameter, input modeling tool obtains the weighing vector of B frequency, B weighing vector iterate got up, obtain comprehensively expecting steering vector.

In technique scheme, described overall treatment module further comprises:

Relevant treatment submodule, obtains and travels through all possible sound source position for described comprehensive reception data vector is done to relevant treatment with comprehensive expectation steering vector, obtains about the decision function of sound source position as follows:

P _coh-conv(θ，r，z)=w _e ^H(θ，r，z)Rw _e(θ，r，z)

Auditory localization module, for by the decision function P of sound source position _coh-convthe estimated value that the location positioning that the maximal value of (θ, r, z) occurs is sound source position.

The invention has the advantages that, in doing single-frequency point normalization, the data after choosing wave beam and forming as a reference, have so just improved the signal to noise ratio (S/N ratio) of reference data to greatest extent.In a word, the present invention proposes the treatment loss bringing for overcoming the incoherent processing of existing wide-band-message, the in the situation that of low signal-to-noise ratio, target detection ability declines, and the present invention proposes the relevant mould base Processing Algorithm in a kind of wave beam normalization broadband.Wave beam normalization of the present invention broadband mould base Processing Algorithm, the reference vector normalization after the array signal of each frequency is formed with wave beam, thereby the minimum detectable signal to noise ratio of reduction processor, improve and detect performance.

Accompanying drawing explanation

Fig. 1 does not adopt when of the present invention, the schematic diagram of mould base location core signal disposal route;

Fig. 2 adopts after the present invention, the schematic diagram of mould base location core signal disposal route;

Fig. 3 is the detailed processing flow chart of algorithm of the present invention;

Fig. 4 adopts the present invention, does not adopt when of the present invention the output contrast of the conventional mould base localization method of embodiment;

Fig. 5 adopts the present invention, does not adopt when of the present invention the output contrast of the adaptive mode base localization method of embodiment;

Fig. 6 does not adopt in embodiment when of the present invention, the output effect figure of incoherent mould base localization method.

Fig. 7 does not adopt in embodiment when of the present invention, the output effect figure of relevant mould base localization method.

Fig. 8 adopts in embodiment when of the present invention, the output effect figure of wave beam normalized mode base localization method.

Embodiment

Below in conjunction with accompanying drawing, the present invention is further described.

, by a plurality of nautical receiving sets, formed, establishing actual array element number is N, a coprocessing B frequency, it is x (t) that array element receives signal indication, snap length is L.Described method comprises the steps:

1) by linear array, receive spacing wave, obtain the snap time-domain signal of N array element constantly;

2) time domain data is done to Fast Fourier Transform (FFT), obtain the array data (X of B frequency ₁, X ₂..., X _b), wherein the array data Xi of i frequency is expressed as follows

X _i=[x _i1,x _i2,...,x _iN] ^T

X wherein _ijthe reception data that represent i frequency of j array element, T represents transposition;

3) data of B frequency are done to wave beam and form, obtain the beam data X of i frequency _{i, beam}for

X _i,beam=X _i*e _i

Wherein, e _ithe array orientation compensation vector that represents i frequency, is determined by element position, target azimuth and signal frequency.

Reference signal (X using this beam data as each frequency _1ref, X _2ref..., X _bref),

X _iref=X _i,beam

Array data (X with this reference signal to each frequency ₁, X ₂..., X _b) do normalization, obtain the data after normalization

as shown in the formula

$({\hat{X}}_1,{\hat{X}}_2,...,{\hat{X}}_B)=(\frac{X_1}{{\mathrm{<figure-callout\; id="1"\; label="X"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_{1\mathrm{ref}}},\frac{X_2}{{\mathrm{<figure-callout\; id="2"\; label="X"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_{2\mathrm{ref}}},...,\frac{X_B}{X_{\mathrm{Bref}}})$

4) by the data after B frequency normalization

iterate gets up, the comprehensive reception data vector that to form a length be B*N

if note

${\hat{X}}_1={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N}]}^T$

Wherein

represent the data after the normalization of i frequency of j array element, the normalized vector after expansion

${\mathrm{<figure-callout\; id="1"\; label="X\; e"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">X</figure-callout>}}_e={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N},{\hat{x}}_{21},{\hat{x}}_{22},...,{\hat{x}}_{2N},...,{\hat{x}}_{\mathrm{<figure-callout\; id="1"\; label="B"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">B</figure-callout>}1},{\hat{x}}_{\mathrm{<figure-callout\; id="2"\; label="B"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">B</figure-callout>}2},...,{\hat{x}}_{\mathrm{BN}}]}^T$

5) receive L snap data constantly, obtain respectively the spread vector of each snap by step 2-4, all the comprehensive reception data vector of L snap is designated as X _{e, all};

X _e，all=[X _e1,X _e2,...X _eL]

Step 1 is to receive snap data constantly, need to repeat L time, and step 5 means this.

6) use the comprehensive reception data vector of L snap to form array covariance matrix

$R=E\left[X_{e,\mathrm{all}}{X}_{e,\mathrm{all}}^H\right]---\left(3\right)$

Wherein, E represents to do statistical average, and H represents conjugate transpose.

7) position that traversal sound source may occur is (θ, r, z), and in conjunction with ocean environment parameter, inputs modeling tool software package KRAKEN, can obtain the weighing vector (ω of B frequency ₁(θ, r, z) ^t, ω ₂(θ, r, z) ^t..., ω _b(θ, r, z) ^t), B weighing vector iterate got up, obtaining length is the comprehensive expectation steering vector of B*N:

w _e(θ，r，z)=[ω ₁(θ，r，z)，ω ₂(θ，r，z)，...，ω _B(θ，r，z)] ^T (4)

Grid number corresponding to scope grid division ，She orientation, distance, the degree of depth that this step need to may occur for sound source is N _θ, N _r, N _z, the modeling number of times needing is N _θ* N _r* N _z

8) according to (3) formula and (4) formula, do relevant treatment, travel through all possible sound source position θ, r, z, can obtain the decision function about sound source position

P _coh-conv(θ，r，z)=w _e ^H(θ，r，z)Rw _e(θ，r，z) (5)

P _coh-convthe position that the maximal value of (θ, r, z) occurs is the estimated value of sound source position.

Basic conception of the present invention is as shown in Figure 2: the data iterate of each frequency is got up, form a comprehensive reception battle array vector that receives data, in order to realize the relevant processing between frequency, to each frequency with a reference number normalization, in the present invention in order to improve the signal to noise ratio (S/N ratio) of reference number as far as possible, employing be the data after wave beam.

Described method comprises the steps, as shown in Figure 3:

Step 1: 101 in corresponding diagram 3, by linear array, receive spacing wave, obtain the time-domain signal of N array element; The signal at the frequencies omega place that i nautical receiving set receives can be written as

$x_w^i=S\left(w\right)G^i(r_i,z_i,B_s,Z_s,w)+Q_w^i---\left(6\right)$

Wherein, S (ω) is Source Spectrum, G ⁱ(r _i, z _i, R _s, Z _s, ω) be the Green function between sound source and i nautical receiving set,

noise component for this frequency place.

The signal that N unit hydrophone array receives is

$x_w={[{\mathrm{<figure-callout\; id="1"\; label="x\; w"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">x</figure-callout>}}_w^{},{\mathrm{<figure-callout\; id="2"\; label="x\; w"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">x</figure-callout>}}_w^{},...,x_w^N]}^T---\left(7\right)$

Step 2: 102 in corresponding diagram 1, time domain data is done to Fast Fourier Transform (FFT), obtain the array data (X of an interested B frequency ₁, X ₂..., X _b);

$x\left(t\right)\&DoubleRightArrow;X\left(\mathrm{\&omega;}\right)---\left(8\right)$

Step 3: 103 in corresponding diagram 1, the data of B frequency are done to wave beam and form, obtain the reference signal (X of each frequency _1ref, X _2ref..., X _bref), the array data with this to each frequency is done normalization, obtains the data after normalization

for example, if not before normalization the data vector of frequency w be

$x_w=\left[\begin{array}{c}\left|S\left(w\right){\mathrm{<figure-callout\; id="1"\; label="G\; w"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">G</figure-callout>}}_w^{}\right|e^j(\arg \left(S\left(\mathrm{\&omega;}\right)\right)+\arg \left(G_w^1\right))\\ .\\ .\\ .\\ \left|S\left(w\right)G_w^N\right|e^{j(\arg \left(S\left(w\right)\right)+\arg \left(G_w^N\right))}\end{array}\right]---\left(9\right)$

After normalization, the data vector of frequency w can be expressed as

Step 4: 104 in corresponding diagram 1, the data iterate after B frequency normalization is got up, the comprehensive reception data vector that to form a length be B*N

Step 5: 105 in corresponding diagram 1, receive the data of L snap, obtain the comprehensive reception data vector of each snap;

Step 6: 106 in corresponding diagram 1, use the comprehensive reception data vector of L snap to form array covariance matrix

$R=E\left[X_e{X}_e^H\right]---\left(11\right)$

Step 7: 107 in corresponding diagram 1, the position of supposing sound source is (θ, r, z), obtain the weighing vector of B frequency according to Marine environment modeling, modeling obtains the weighing vector of receiving array, and B weighing vector iterate got up, obtaining length is the comprehensive expectation steering vector of B*N

w _e=[w ₁，w ₂，...，w _B] ^T (12)

$w_i=\frac{{[G({\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">r</figure-callout>}}_1,{\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">z</figure-callout>}}_1,{\mathrm{\&theta;}}_s,r_s,z_s),G({\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">r</figure-callout>}}_2,{\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">z</figure-callout>}}_2,{\mathrm{\&theta;}}_s,r_s,z_s),...,G(r_N,z_N,{\mathrm{\&theta;}}_s,r_s,z_s)]}^T}{\left|\right[G({\mathrm{<figure-callout\; id="1"\; label="r"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">r</figure-callout>}}_1,{\mathrm{<figure-callout\; id="1"\; label="z"\; filenames="HDA00001789102100042.png,HDA00001789102100051.png"\; state="\{\{state\}\}">z</figure-callout>}}_1,{\mathrm{\&theta;}}_s,r_s,z_s),G({\mathrm{<figure-callout\; id="2"\; label="r"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">r</figure-callout>}}_2,{\mathrm{<figure-callout\; id="2"\; label="z"\; filenames="HDA00001789102100051.png"\; state="\{\{state\}\}">z</figure-callout>}}_2,{\mathrm{\&theta;}}_s,r_s,z_s),...,G(r_N,z_N,{\mathrm{\&theta;}}_s,r_s,z_s)\left]\right|}---\left(13\right)$

Step 8: 108 in corresponding diagram 1, according to (11) formula and (12) formula, use normal signal disposal route, travel through all possible sound source position θ, r, z, obtains the decision function about target location

P _coh-conv(θ，r，z)=w _e ^H(θ，r，z)Rw _e(θ，r，z) (14)

Use steering vector adaptive optimization technology, travel through all possible sound source position θ, r, z, the decision function that obtains adaptive signal processing method Shi， target location is:

$P_{\mathrm{coh}-\mathrm{mvdr}}(\mathrm{\&theta;},r,z)=\frac{1}{{w_e}^H(\mathrm{\&theta;},r,z)R^{-1}{w}_e(\mathrm{\&theta;},r,z)}---\left(15\right)$

Above-mentioned P _coh-conv(θ, r, z), P _coh-mvdrthe position that the maximal value of (θ, r, z) occurs is the estimated value of sound source position

In addition, not being both of above-mentioned two kinds of decision functions determined by Fig. 2 (304) step, if (304) step is not taked Optimized Measures, obtains P _coh-conv(θ, r, z), if (304) are optimized according to certain Optimality Criteria and constraint condition (such as maximum array gain, optimum signal to noise ratio (S/N ratio) or robust adaptive technology etc.), obtains P _coh-mvdr(θ, r, z).

Embodiment

Below in conjunction with data simulation and accompanying drawing, the specific embodiment of the present invention is described in further detail.

Simulated conditions: 10 yuan of basic matrixs are positioned at seabed, array element distance 5m, gets typical shallow water environment, and the dark 30m in sea, waits sound velocity gradient, and sound source frequency band 400-600Hz, gets 5 frequencies wherein.While obtaining using respectively conventional treatment method and MVDR disposal route, Wideband Incoherent disposal route (inc), coherent signal subspace method (MP), improved coherent signal subspace method (MP-beam) peak value/background ratio are with the change curve of different input signal-to-noise ratios, as shown in Fig. 4-Fig. 5.

Can see, the reduction with folding signal to noise ratio (S/N ratio), adopts advantage of the present invention more obvious.The result of using normal signal disposal route (not doing 304 optimization), as shown in Figure 4, during high s/n ratio, performance of the present invention is between two kinds of classic methods, and during low signal-to-noise ratio, performance of the present invention is better than two kinds of classic methods.The situation of adaptive signal processing method, as shown in Figure 5, under various state of signal-to-noise, performance of the present invention is better than two kinds of classic methods all the time.

Fig. 6-Fig. 8 has provided in input signal-to-noise ratio-5dB situation, and the Output rusults of three kinds of adaptive mode base localization methods, can see, result of the present invention (Fig. 8) is obviously better than two kinds of classic methods.

In a word, the present invention, than method and the relevant method of processing of array element normalization of traditional incoherent processing, has more excellent detection performance.

It should be noted last that, above embodiment is only unrestricted in order to technical scheme of the present invention to be described.Although the present invention is had been described in detail with reference to embodiment, those of ordinary skill in the art is to be understood that, technical scheme of the present invention is modified or is equal to replacement, do not depart from the spirit and scope of technical solution of the present invention, it all should be encompassed in the middle of claim scope of the present invention.

Claims (10)

1. The 1.Yi Zhong broadband mould base signal processing method that is concerned with, described method comprises:

   Step 101) obtain the array received data of B frequency, the data of Whole frequency band are done to wave beam normalization, relevant formation comprehensively receives data vector, comprised the reception data vector of Whole frequency band information;

   Step 102) obtain the modeling vector of B frequency, and the modeling vector of Whole frequency band is concerned with and forms comprehensive expectation steering vector, comprised the expectation steering vector of Whole frequency band information;

   Step 103) according to described comprehensive reception data vector and comprehensive expectation steering vector, obtain optimum steering vector;

   Step 104) according to optimum steering vector, obtain final target localization result.
2. 2. the relevant mould base signal processing method in broadband according to claim 1, is characterized in that described step 101) be specially: the array received data to B the frequency receiving are carried out Fast Fourier Transform (FFT), obtain the data of each single-frequency point; The data of B frequency are carried out to wave beam formation and normalized successively, form the comprehensive data vector that receives; The array received data configuration array covariance matrix of L snap of accumulation.
3. 3. the relevant mould base signal processing method in broadband according to claim 1, is characterized in that described step 101) further comprise following sub-step:

   Step 101-1) by linear array, receive spacing wave, obtain the snap time-domain signal of N array element constantly;

   Step 101-2) time domain data is done to Fast Fourier Transform (FFT), obtain the array data (X of B frequency ₁, X ₂..., X _b), wherein the array data Xi of i frequency is expressed as follows

   X _i=[x _i1,x _i2,...,x _iN] ^T

   X wherein _ijthe reception data that represent i frequency of j array element, T represents transposition;

   Step 101-3) data of B frequency are done to wave beam and form, obtain the beam data X of i frequency _{i, beam}for

   X _i,beam=C _i*e _i

   Wherein, e _ithe array orientation compensation vector that represents i frequency, is determined by element position, target azimuth and signal frequency;

   Reference signal (X using this beam data as each frequency _1ref, X _2ref..., X _bref),

   X _iref=X _i,beam

   Array data (X with this reference signal to each frequency ₁, X ₂..., X _b) do normalization, obtain the data after normalization

   

   as shown in the formula:

   $({\hat{X}}_1,{\hat{X}}_2,...,{\hat{X}}_B)=(\frac{X_1}{X_{1\mathrm{ref}}},\frac{X_2}{X_{2\mathrm{ref}}},...,\frac{X_B}{X_{\mathrm{Bref}}})$

   Step 101-4) by the data after B frequency normalization

   

   iterate gets up, the comprehensive reception data vector that to form a length be B*N

   

   if note:

   ${\hat{X}}_1={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N}]}^T$

   Wherein

   

   represent the data after the normalization of i frequency of j array element, the normalized vector after expansion:

   $X_{e1}={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N},{\hat{x}}_{21},{\hat{x}}_{22},...,{\hat{x}}_{2N},...,{\hat{x}}_{B1},{\hat{x}}_{B2},...,{\hat{x}}_{\mathrm{BN}}]}^T;$

   Step 101-5) receive L snap data constantly, respectively by step 101-2)-101-4) obtain the comprehensive reception data vector of each snap, the comprehensive reception data vector of whole L snaps is designated as X _{e, all};

   X _e，all=[X _e1,X _e2,...X _eL]

   Described step 101-1) be to receive snap data constantly, need to repeat L time step 101-5) mean this;

   Step 101-6) use the comprehensive reception data vector of L snap to form array covariance matrix

   $R=E\left[X_{e,\mathrm{all}}{X}_{e,\mathrm{all}}^H\right]---\left(1\right)$

   Wherein, E represents to do statistical average, and H represents conjugate transpose.
4. 4. the relevant mould base signal processing method in broadband according to claim 1; it is characterized in that; described step 102) concrete steps are: the position that traversal sound source may occur; and in conjunction with ocean environment parameter; input modeling tool obtains the weighing vector of B frequency; B weighing vector iterate got up, obtain comprehensively expecting steering vector.
5. 5. the relevant mould base signal processing method in broadband according to claim 1, is characterized in that described step 103) further comprise:

   Step 103-1) described comprehensive reception data vector and comprehensive expectation steering vector are done to the decision function value that relevant treatment obtains this sound source position, and travel through all possible sound source position, obtain about the decision function of whole sound source positions as follows:

   P _coh-conv(θ，r，z)=w _e ^H(θ，r，z)Rw _e(θ，r，z)

   The decision function P of step 103-2 sound source position _coh-convthe position that the maximal value of (θ, r, z) occurs is the estimated value of sound source position.
6. The 6.Yi Zhong broadband mould base signal processing system that is concerned with, described system comprises:

   The comprehensive data vector acquisition module that receives, for obtaining the array received data of B frequency, does wave beam normalization to the data of Whole frequency band, and relevant formation comprehensively receives data vector;

   Comprehensive expectation steering vector acquisition module, for obtaining the modeling vector of B frequency, and is concerned with and forms comprehensive expectation steering vector the modeling vector of Whole frequency band;

   Overall treatment module, for the comprehensive reception data vector according to described and comprehensive expectation steering vector, obtains optimum steering vector; And obtain final target localization result according to optimum steering vector.
7. 7. the relevant mould base signal processing system in broadband according to claim 6, is characterized in that, described comprehensive reception data vector acquisition module carries out Fast Fourier Transform (FFT) to the array received data of B the frequency receiving, and obtains the data of each single-frequency point; The data of B frequency are carried out to wave beam formation and normalized successively, form the comprehensive data vector that receives; Accumulate L snap array received data configuration array covariance matrix.
8. 8. the relevant mould base signal processing system in broadband according to claim 6, is characterized in that, described comprehensive reception data vector acquisition module further comprises:

   Receive submodule, for receiving spacing wave by linear array, obtain the snap time-domain signal of N array element constantly;

   First processes submodule, and time domain data is done to Fast Fourier Transform (FFT), obtains the array data (X of B frequency ₁, X ₂..., X _b), wherein the array data Xi of i frequency is expressed as follows

   X _i=[x _i1,x _i2,...,x _iN] ^T

   X wherein _ijthe reception data that represent i frequency of j array element, T represents transposition;

   Second processes submodule, for the data of B frequency being done to wave beam, forms, and obtains the beam data X of i frequency _{i, beam}for

   X _i，beam=X _i*e _i

   Wherein, e _ithe array orientation compensation vector that represents i frequency, is determined by element position, target azimuth and signal frequency;

   Reference signal (X using this beam data as each frequency _1ref, X _2ref..., X _bref),

   X _iref=X _i,beam

   Array data (X with this reference signal to each frequency ₁, X ₂..., X _b) do normalization, obtain the data after normalization as shown in the formula:

   $({\hat{X}}_1,{\hat{X}}_2,...,{\hat{X}}_B)=(\frac{X_1}{X_{1\mathrm{ref}}},\frac{X_2}{X_{2\mathrm{ref}}},...,\frac{X_B}{X_{\mathrm{Bref}}})$

   Iterate submodule, for by the data after B frequency normalization iterate gets up, the comprehensive reception data vector that to form a length be B*N

   

   if note:

   ${\hat{X}}_1={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N}]}^T$

   Wherein

   

   represent the data after the normalization of i frequency of j array element, the normalized vector after expansion:

   $X_{e1}={[{\hat{x}}_{11},{\hat{x}}_{12},...,{\hat{x}}_{1N},{\hat{x}}_{21},{\hat{x}}_{22},...,{\hat{x}}_{2N},...,{\hat{x}}_{B1},{\hat{x}}_{B2},...,{\hat{x}}_{\mathrm{BN}}]}^T;$

   The 3rd processes submodule, for receiving L snap data constantly, respectively by step 101-2)-101-4) obtain the comprehensive reception data vector of each snap, the comprehensive reception data vector of whole L snaps is designated as X _{e, all};

   X _e，all=[X _e1,X _e2,...X _eL]

   Described step 101-1) be to receive snap data constantly, need to repeat L time step 101-5) mean this;

   Array covariance matrix forms submodule, for using the comprehensive reception data vector of L snap to form array covariance matrix

   $R=E\left[X_{e,\mathrm{all}}{X}_{e,\mathrm{all}}^H\right]---\left(2\right)$

   Wherein, E represents to do statistical average, and H represents conjugate transpose.
9. 9. the relevant mould base signal processing system in broadband according to claim 6; it is characterized in that; the position that described comprehensive expectation steering vector acquisition module traversal sound source may occur; and in conjunction with ocean environment parameter; input modeling tool obtains the weighing vector of B frequency; B weighing vector iterate got up, obtain comprehensively expecting steering vector.
10. 10. the relevant mould base signal processing system in broadband according to claim 6, is characterized in that, described overall treatment module further comprises:

    Relevant treatment submodule, obtains and travels through all possible sound source position for described comprehensive reception data vector is done to relevant treatment with comprehensive expectation steering vector, obtains about the decision function of sound source position as follows:

    P _coh-conv(θ，r，z)=w _e ^H(θ，r，z)Rw _e(θ，r，z)

    Auditory localization module, for by the decision function P of sound source position _coh-convthe estimated value that the location positioning that the maximal value of (θ, r, z) occurs is sound source position.

Images